Darsh S.

I build agentic AI and data systems that run in production not demos, not POCs that die in staging. My stack: multi-agent orchestration, RAG pipelines, tool calling, memory management, MCP servers, and end-to-end deployment on Cloud. I have shipped Data systems serving 10,000+ users, handling millions of daily queries, and running continuously for 2+ years. What I bring that most AI engineers don't: I have sat in the room with clients when production breaks, diagnosed it, and fixed it. I architect, I ship, and I stay accountable for outcomes not just delivery. Current focus areas: - Multi-agent architectures: LangGraph, CrewAI, Google ADK, MCP - RAG systems: vector search, hybrid retrieval, cross-encoder reranking - Google Cloud: Vertex AI, Agent Runtime, BigQuery, Cloud Run, Pub/Sub - LLMs: Gemini, Claude, Llama - whichever fits the production constraint, Fine-Tuning I also run devlit.ai, a technical learning platform on production AI systems. Which means I explain complex architectures clearly, not just build them. GCP Professional Data Engineer · GCP Professional Cloud Architect